On Tue, Nov 5, 2013 at 1:41 PM, Mike Rylander <mrylan...@gmail.com> wrote: > CORRECTION: the commit hash is > 3e7f6f1503938e3e5a1cfe3bd90eaa8f021a0d48 ... I failed to fetch before > attempting to pick. Sorry!
Hey Mike: It looks like that test assumes that the concerto sample records have been loaded, which means we can't run it on a production system. What I've done in my tests, instead, is insert the desired record with a known-and-entirely-improbable record ID, then tested against that. As we're using an explicit ID, the sequence doesn't get touched, and as the test occurs within a transaction, the record and all derived rows get rolled back at the end. To make my approach more airtight, I suppose I should test for the existence of a record with the entirely-improbable ID first, and skip the test if that record ID for some reason exists...